R Markdown file

library("tidyverse"); theme_set(theme_bw())
## ── Attaching packages ─────────────────────────────────────── tidyverse 1.3.2 ──
## ✔ ggplot2 3.4.0      ✔ purrr   0.3.4 
## ✔ tibble  3.1.8      ✔ dplyr   1.0.10
## ✔ tidyr   1.2.1      ✔ stringr 1.4.1 
## ✔ readr   2.1.2      ✔ forcats 0.5.2 
## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag()    masks stats::lag()
library("Sleuth3")

# Tables
library("knitr") # for kable
library("kableExtra")
## Warning: package 'kableExtra' was built under R version 4.2.3
## 
## Attaching package: 'kableExtra'
## 
## The following object is masked from 'package:dplyr':
## 
##     group_rows
library("formattable")
## Warning: package 'formattable' was built under R version 4.2.3
library("DT")
## Warning: package 'DT' was built under R version 4.2.3
# Figures
library("scales")
## 
## Attaching package: 'scales'
## 
## The following objects are masked from 'package:formattable':
## 
##     comma, percent, scientific
## 
## The following object is masked from 'package:purrr':
## 
##     discard
## 
## The following object is masked from 'package:readr':
## 
##     col_factor
library("plotly")
## 
## Attaching package: 'plotly'
## 
## The following object is masked from 'package:formattable':
## 
##     style
## 
## The following object is masked from 'package:ggplot2':
## 
##     last_plot
## 
## The following object is masked from 'package:stats':
## 
##     filter
## 
## The following object is masked from 'package:graphics':
## 
##     layout
library("leaflet")
## Warning: package 'leaflet' was built under R version 4.2.3
library("gifski")
## Warning: package 'gifski' was built under R version 4.2.3

Rmarkdown documents that produce HTML files can include a variety of features that provide an interactive document for the user. Primarily this interactivity is implemented as will concern stand-alone tables, figures, and animations (movies). Typically this interactivity is available via an R package interface to a javascript library.

We’ll take a look at the construction of tables using the knitr, formattable, and DT packages. Technically, the first two packages provide non-interactive tables while the third provides interactivity. But we’ll start with the first two as they provide some nice functionality to make nice looking HTML tables.

Tables

We will take a look at the diamonds data set.

dim(diamonds)
## [1] 53940    10

These data are too large for interactive scatterplots and thus we will take a random sample of these data.

kable

The kable() function in the knitr package provides an easy display of tables in an HTML document.

By default, the kable function will show the entire table. So, let’s just show the first few lines.

d <- diamonds %>%
  group_by(cut) %>% # ensure we have all cuts for grouping
  sample_n(3)

Also, by default, the table looks pretty bad, so let’s add some styling.

knitr::kable(d) %>% 
  kable_styling()
carat cut color clarity depth table price x y z
1.00 Fair D SI1 67.3 57 5096 6.15 6.04 4.10
2.50 Fair J SI2 66.9 58 14652 8.36 8.22 5.55
2.20 Fair I I1 66.3 56 7934 8.05 8.00 5.32
2.01 Good G SI2 63.6 59 15832 7.95 8.00 5.07
1.00 Good E VS1 63.6 59 7260 6.32 6.35 4.03
1.01 Good D VS2 64.0 58 6176 6.30 6.24 4.01
0.44 Very Good I VVS1 60.8 57 926 4.88 4.93 2.98
0.51 Very Good D VS2 60.5 60 1787 5.16 5.19 3.13
1.00 Very Good E VS2 62.8 63 5720 6.28 6.33 3.96
0.73 Premium I VVS2 61.5 58 2519 5.78 5.73 3.54
1.20 Premium G VS1 62.5 59 8056 6.72 6.64 4.19
0.43 Premium F SI1 61.3 60 1016 4.86 4.83 2.97
1.01 Ideal F VS1 62.0 57 7745 6.42 6.46 3.99
0.41 Ideal G IF 61.7 56 1259 4.78 4.81 2.96
0.33 Ideal F VVS1 61.8 53 880 4.48 4.51 2.78

Formatting

d %>%
knitr::kable(
  caption = "Diamonds data", 
  align = c("rlllrrrrrr")
) %>% 
  kable_styling(bootstrap_options = c('striped', 'hover', 'condensed')) %>%
  scroll_box(height = "200px")
Diamonds data
carat cut color clarity depth table price x y z
1.00 Fair D SI1 67.3 57 5096 6.15 6.04 4.10
2.50 Fair J SI2 66.9 58 14652 8.36 8.22 5.55
2.20 Fair I I1 66.3 56 7934 8.05 8.00 5.32
2.01 Good G SI2 63.6 59 15832 7.95 8.00 5.07
1.00 Good E VS1 63.6 59 7260 6.32 6.35 4.03
1.01 Good D VS2 64.0 58 6176 6.30 6.24 4.01
0.44 Very Good I VVS1 60.8 57 926 4.88 4.93 2.98
0.51 Very Good D VS2 60.5 60 1787 5.16 5.19 3.13
1.00 Very Good E VS2 62.8 63 5720 6.28 6.33 3.96
0.73 Premium I VVS2 61.5 58 2519 5.78 5.73 3.54
1.20 Premium G VS1 62.5 59 8056 6.72 6.64 4.19
0.43 Premium F SI1 61.3 60 1016 4.86 4.83 2.97
1.01 Ideal F VS1 62.0 57 7745 6.42 6.46 3.99
0.41 Ideal G IF 61.7 56 1259 4.78 4.81 2.96
0.33 Ideal F VVS1 61.8 53 880 4.48 4.51 2.78

Grouping

groups <- table(d$cut)

d %>%
knitr::kable(
  caption = "Diamonds data", 
  align = c("rlllrrrrrr")
) %>% 
  kable_styling(bootstrap_options = c('striped', 'hover', 'condensed')) %>%
  pack_rows(
    index = setNames(groups, names(groups))
  )
Diamonds data
carat cut color clarity depth table price x y z
Fair
1.00 Fair D SI1 67.3 57 5096 6.15 6.04 4.10
2.50 Fair J SI2 66.9 58 14652 8.36 8.22 5.55
2.20 Fair I I1 66.3 56 7934 8.05 8.00 5.32
Good
2.01 Good G SI2 63.6 59 15832 7.95 8.00 5.07
1.00 Good E VS1 63.6 59 7260 6.32 6.35 4.03
1.01 Good D VS2 64.0 58 6176 6.30 6.24 4.01
Very Good
0.44 Very Good I VVS1 60.8 57 926 4.88 4.93 2.98
0.51 Very Good D VS2 60.5 60 1787 5.16 5.19 3.13
1.00 Very Good E VS2 62.8 63 5720 6.28 6.33 3.96
Premium
0.73 Premium I VVS2 61.5 58 2519 5.78 5.73 3.54
1.20 Premium G VS1 62.5 59 8056 6.72 6.64 4.19
0.43 Premium F SI1 61.3 60 1016 4.86 4.83 2.97
Ideal
1.01 Ideal F VS1 62.0 57 7745 6.42 6.46 3.99
0.41 Ideal G IF 61.7 56 1259 4.78 4.81 2.96
0.33 Ideal F VVS1 61.8 53 880 4.48 4.51 2.78

Highlighting

d %>%
  # Conditional highlighting
  mutate(
    carat = cell_spec(carat, "html", color = ifelse(carat > .7, "red", "black")),
    price = cell_spec(price, "html", color = ifelse(price < 5000, "blue", "black"))
  ) %>%
  
knitr::kable(
  escape = FALSE,
  caption = "Diamonds data", 
  align = c("rlllrrrrrr")
) %>% 
  kable_styling(bootstrap_options = c('striped', 'hover', 'condensed'))
Diamonds data
carat cut color clarity depth table price x y z
1 Fair D SI1 67.3 57 5096 6.15 6.04 4.10
2.5 Fair J SI2 66.9 58 14652 8.36 8.22 5.55
2.2 Fair I I1 66.3 56 7934 8.05 8.00 5.32
2.01 Good G SI2 63.6 59 15832 7.95 8.00 5.07
1 Good E VS1 63.6 59 7260 6.32 6.35 4.03
1.01 Good D VS2 64.0 58 6176 6.30 6.24 4.01
0.44 Very Good I VVS1 60.8 57 926 4.88 4.93 2.98
0.51 Very Good D VS2 60.5 60 1787 5.16 5.19 3.13
1 Very Good E VS2 62.8 63 5720 6.28 6.33 3.96
0.73 Premium I VVS2 61.5 58 2519 5.78 5.73 3.54
1.2 Premium G VS1 62.5 59 8056 6.72 6.64 4.19
0.43 Premium F SI1 61.3 60 1016 4.86 4.83 2.97
1.01 Ideal F VS1 62.0 57 7745 6.42 6.46 3.99
0.41 Ideal G IF 61.7 56 1259 4.78 4.81 2.96
0.33 Ideal F VVS1 61.8 53 880 4.48 4.51 2.78

formattable

Another function is formattable() in the formattable package. The default table is reasonable.

d %>%
  formattable::formattable()
carat cut color clarity depth table price x y z
1.00 Fair D SI1 67.3 57 5096 6.15 6.04 4.10
2.50 Fair J SI2 66.9 58 14652 8.36 8.22 5.55
2.20 Fair I I1 66.3 56 7934 8.05 8.00 5.32
2.01 Good G SI2 63.6 59 15832 7.95 8.00 5.07
1.00 Good E VS1 63.6 59 7260 6.32 6.35 4.03
1.01 Good D VS2 64.0 58 6176 6.30 6.24 4.01
0.44 Very Good I VVS1 60.8 57 926 4.88 4.93 2.98
0.51 Very Good D VS2 60.5 60 1787 5.16 5.19 3.13
1.00 Very Good E VS2 62.8 63 5720 6.28 6.33 3.96
0.73 Premium I VVS2 61.5 58 2519 5.78 5.73 3.54
1.20 Premium G VS1 62.5 59 8056 6.72 6.64 4.19
0.43 Premium F SI1 61.3 60 1016 4.86 4.83 2.97
1.01 Ideal F VS1 62.0 57 7745 6.42 6.46 3.99
0.41 Ideal G IF 61.7 56 1259 4.78 4.81 2.96
0.33 Ideal F VVS1 61.8 53 880 4.48 4.51 2.78 
d %>%
  
  # Conditional highlighting
  mutate(
    carat = cell_spec(carat, "html", color = ifelse(carat > .7, "red", "black")),
    price = cell_spec(price, "html", color = ifelse(price < 5000, "blue", "black"))
  ) %>%
  
  formattable::formattable(
    list(
      # Width depends on proportion from 0 to max value
      x = color_bar("#C8102E"),    
      y = color_bar("#C8102E"),    
      z = color_bar("#C8102E"),    
      
      # Color depends on proportion from min to max value
      depth = color_tile("#CAC7A7","#524727")
    )
  )
carat cut color clarity depth table price x y z
1 Fair D SI1 67.3 57 5096 6.15 6.04 4.10
2.5 Fair J SI2 66.9 58 14652 8.36 8.22 5.55
2.2 Fair I I1 66.3 56 7934 8.05 8.00 5.32
2.01 Good G SI2 63.6 59 15832 7.95 8.00 5.07
1 Good E VS1 63.6 59 7260 6.32 6.35 4.03
1.01 Good D VS2 64.0 58 6176 6.30 6.24 4.01
0.44 Very Good I VVS1 60.8 57 926 4.88 4.93 2.98
0.51 Very Good D VS2 60.5 60 1787 5.16 5.19 3.13
1 Very Good E VS2 62.8 63 5720 6.28 6.33 3.96
0.73 Premium I VVS2 61.5 58 2519 5.78 5.73 3.54
1.2 Premium G VS1 62.5 59 8056 6.72 6.64 4.19
0.43 Premium F SI1 61.3 60 1016 4.86 4.83 2.97
1.01 Ideal F VS1 62.0 57 7745 6.42 6.46 3.99
0.41 Ideal G IF 61.7 56 1259 4.78 4.81 2.96
0.33 Ideal F VVS1 61.8 53 880 4.48 4.51 2.78

DT

As we will see, with the pagination, datatable() provides the capability to succinctly display much larger tables. So we will use more data

set.seed(20230416)
d <- diamonds %>%
  sample_n(1000)

A basic interactive table using DT::datatable().

DT::datatable(d)

Many options can be added

Filtering

DT::datatable(d, rownames = FALSE, filter = "top")

Buttons

DT::datatable(d, rownames = FALSE, 
              extensions = "Buttons",
              options = list(
                dom = "Bfrtip",
                buttons = c("copy","csv","excel","pdf","print")
              ))

Editing

DT::datatable(d, rownames = FALSE, 
              editable = TRUE,
              extensions = "Buttons",
              options = list(
                dom = "Bfrtip",
                buttons = c("copy","csv","excel","pdf","print")
              ))

Figures

In this section, I am combining graphics, i.e. plots, as well as maps and animations (movies).

Plots

There are a variety of approaches to including interactivity in graphics in rmarkdown documents. We’ll focus on using the plotly library and specifically the ggplotly() function which provides interactivity for ggplot2 created graphics.

plotly::ggplotly()

The ggplotly() function from the plotly package provides interactivity for (all?) ggplot2 constructed graphics. The interactivity provide allows the user to

  • resize (zoom, rescale, reset)
  • pan
  • hover (show vs compare)
  • toggle spike lines
  • download

Boxplot

g <- ggplot(case0501, aes(x = Diet, y = Lifetime)) + 
  geom_boxplot() +
  coord_flip()

ggplotly(g)

Histogram

g <- ggplot(diamonds, aes(x = price)) + 
  geom_histogram(bins = 100)

ggplotly(g)

Scatterplot

Here is a static plot of the diamonds data set.

d <- diamonds %>% sample_n(1000)

g <- ggplot(d, 
            aes(
              x = carat, 
              y = price,
              shape = cut,
              color = color)) + 
  geom_point() +
  scale_y_log10() + 
  scale_x_log10(breaks = scales::breaks_pretty()) 

g
## Warning: Using shapes for an ordinal variable is not advised

ggplotly(g)
## Warning: Using shapes for an ordinal variable is not advised

It seems plotly.js does not support multiple legends.

dygraphs()

Another package from constructing interactive graphics is dygraphs.

Maps

leaflet()

Example taken from here.

leaflet::leaflet() %>% 
  addTiles() %>%
  setView(-93.65, 42.0285, zoom = 17) %>%
  addPopups(
    -93.65, 42.0285,
    'Here is the <b>Department of Statistics</b>, ISU'
  )

Animations

gibbs_bivariate_normal = function(theta0, n_points, rho) {
  theta = matrix(theta0, nrow=n_points, ncol=2, byrow=TRUE)
  v = sqrt(1-rho^2)
  for (i in 2:n_points) {
    theta[i,1] = rnorm(1, rho*theta[i-1,2], v)
    theta[i,2] = rnorm(1, rho*theta[i  ,1], v)
  }
  return(theta)
}

theta = gibbs_bivariate_normal(c(-3,3), n<-20, rho=rho<-0.9)
bivariate_normal_animation = function(x, rho, ask=interactive()) {
  # Create contour plot
  n.out = 101
  xx <- seq(-3, 3, length=n.out)
  grid <- expand.grid(x=xx, y=xx)
  Sigma = diag(rep(.1,2))+rho
  like <- matrix(apply(grid, 1, function(x) mvtnorm::dmvnorm(x,sigma=Sigma)),n.out,n.out)
  
  for (i in 2:nrow(x)) {
    jj = (2:i)[-(i-1)] # vector from 2:(i-1) and NULL if i=2
    for (j in 1:6) {
      plot.new()
      
      # All previous plotting
      contour(xx, xx, like, drawlabels=F, nlevels=10, xlim=c(-3,3), ylim=c(-3,3), 
              xlab=expression(theta[1]), ylab=expression(theta[2]))  
      segments(x[jj-1,1], x[jj-1,2], x[jj,1], x[jj-1,2], col="gray")
      segments(x[jj  ,1], x[jj-1,2], x[jj,1], x[jj  ,2], col="gray")
      points(x[(1:(i-1)),1], x[(1:(i-1)),2], col="red", pch=19)
      
      # New plotting
      if (j>1 & j<4) abline(h=x[i-1,2], lty=2)
      if (j>2) arrows(x[i-1,1], x[i-1,2], x[i,1], x[i-1,2], length=0.1)
      if (j>3 & j<6) abline(v=x[i,1], lty=2)
      if (j>4) arrows(x[i,1], x[i-1,2], x[i,1], x[i,2], length=0.1)
      if (j>5) points(x[i,1], x[i,2], col="red", pch=19)
      
      if (ask) readline("hit <enter>:")
    }
  }
  
  jj=2:nrow(x)
  contour(xx, xx, like, drawlabels=F, nlevels=10, xlim=c(-3,3), ylim=c(-3,3), 
          xlab=expression(theta[1]), ylab=expression(theta[2]))  
  segments(x[jj-1,1], x[jj-1,2], x[jj,1], x[jj-1,2], col="gray")
  segments(x[jj  ,1], x[jj-1,2], x[jj,1], x[jj  ,2], col="gray")
  points(x[,1], x[,2], col="red", pch=19)
}
bivariate_normal_animation(theta, rho = 0.9)

Additional resources

Galleries

Official:

Individual sites

Individuals:

Embed

You can always embed additional interactivity. To get this to work, you need to add the option data-external="1" to the iframe options.

For example, here is a google map.

Here is an embedded video of mine from YouTube discussing the Gibbs sampler demonstrated above.